Pentest Api Attacker

Security checks across malware telemetry and agentic risk

Overview

This skill has authorization safeguards, but it appears to generate realistic security findings without performing the advertised API testing.

Install only if you treat this as a scaffold or demo, not a working API pentest tool. Do not rely on its generated findings for audit evidence, remediation decisions, or assurance until real test execution is implemented and reports clearly distinguish placeholder data from verified results.

SkillSpector

By NVIDIA

Vulnerability Patterns

MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (3)

Description-Behavior Mismatch

High

Confidence: 99% confidence
Finding: This is a true integrity/security issue: the skill advertises active API security testing, but in reality it only checks scope/authorization flags and emits placeholder artifacts. In a pentest context, producing synthetic findings and reports as if testing occurred can mislead operators into believing an assessment was performed, causing false assurance, wasted remediation effort, or incorrect security decisions.

Intent-Code Divergence

Medium

Confidence: 96% confidence
Finding: The documentation and CLI description materially misrepresent the capability of the tool, which is dangerous in a security-testing skill because users may rely on it for assurance or evidence of assessment coverage. While it does not directly exploit a target, it undermines trust and can lead to missed vulnerabilities because no real API enumeration or attack logic exists.

Intent-Code Divergence

High

Confidence: 99% confidence
Finding: The generated finding falsely states that automated and manual testing identified a security weakness, despite no testing being performed before report generation. In a pentesting workflow, fabricated findings are dangerous because they can trigger unnecessary incident response or remediation, while also obscuring the fact that the target was never actually assessed.

VirusTotal

66/66 vendors flagged this skill as clean.

View on VirusTotal